Accepty launches new products to the accelerate growth
BERLIN—Accepty, the programmable financial services company, today showcased new products to accelerate the growth of German


Today we're announcing our first set of Al agents, built and trained to think like your sharpest controller and work like a thousand of them-around the clock. Picture the most routine transaction in your company: Jess from sales grabs a $5 latte on the way to a client meeting.
1. She saves the receipt and uploads a picture when she gets home: 4 minutes.
2. Closed-loop feedback. Every decision is logged, evaluated, and used to improve the model. This is software that learns the way your team thinks.
3. A finance associate opens NetSuite, audits policy, codes the expense, syncs the entry: another 7 minutes.
agents flip that equation. These are the first autonomous finance agents purpose- built for real-world policy enforcement, fraud detection, and policy improvement. Give the agent your PDF policy and it immediately starts approving low-risk expenses, flagging true outliers, answering employee questions over SMS, and suggesting policy improvements.
Starting today, Ramp agents are available for Ramp Plus users. Ramp agents act like autonomous, tireless reviewers, catching 15x more out-of-policy spend than non-Al alternatives and enforcing policy with 99% accuracy. Agents escalate only the 10-15% of expenses that need further human judgment. The result: higher precision, immediate savings, and dramatically better ROI than human review.
Meet our first set of agents, designed for controllers
Agents immediately start working for you. No separate systems. No new logins. Just intelligence layered directly into the workflows and integrations you already use with Ramp. Here's how it works:
1. Written policy + real-world decisions. agents ingest both the written rules and the messy exceptions hidden in historical approvals.
2. Full auditability. Every action comes with a rationale. If the agent made the call, you know why and you can trace it back-every time.
3. Closed-loop feedback. Every decision is logged, evaluated, and used to improve the model. This is software that learns the way your team thinks.

